2024 was the year the AI and blockchain industries stopped being narrated as rivals and started being treated as complementary stacks. The shift didn’t happen because of any single announcement — it happened because the practical questions on real product teams (“how does my agent transact?”, “how do I prove this output is mine?”, “how do I keep the data inside the building?”) kept arriving at the same intersection.
The AI year
- Agent frameworks matured. Tooling for building production-grade AI agents (LangChain, LlamaIndex, AutoGen, OpenAI Assistants, Anthropic’s Claude with tool use, custom in-house stacks) moved from experimental to deployable.
- The Model Context Protocol (MCP) emerged. Anthropic’s open standard for connecting AI assistants to data sources and tools quickly became a de facto integration surface that other vendors started supporting. We built our own Ethora MCP server on top of it.
- Multimodal became the default. Text-only systems became the exception rather than the rule.
- Enterprise compliance momentum. SOC 2-ready, HIPAA-aware, EU AI Act-aligned offerings became a genuine commercial category rather than a marketing afterthought.
The blockchain year
- Real-world asset tokenization gained real volume. Treasuries, money market funds, private credit — not headline-grabbing for retail, but quietly load-bearing for the next institutional cycle.
- Layer 2 maturity. Base, Arbitrum, Optimism, Polygon zkEVM crossed the threshold from “scaling experiment” to “where serious dev teams ship by default.”
- Stablecoin growth held. Stablecoins continued to settle ever-larger volumes globally, and were increasingly used as a unit of account for non-crypto-native applications.
- Regulatory progress in specific places. The EU’s MiCA framework went into effect, the US started clarifying its position post-election, and the UK kept inching toward operational clarity.
Where the two industries actually met
The convergence themes that mattered in 2024:
- AI agents with wallets. Account abstraction (ERC-4337) matured enough that an autonomous agent owning and operating its own wallet went from a slide deck to a working pattern.
- On-chain content provenance for AI output. As generative content became indistinguishable from human-made by inspection, on-chain attestations of “this was produced by model M at time T from inputs X” started appearing in publishing, news, and IP workflows.
- Decentralized identity for agents. Verifiable credentials and DIDs got applied to AI agents — proving an agent’s affiliation, capability, and authorisation in machine-readable form. We touched on this overlap earlier and it now feels less speculative.
- Compute markets and DePIN. Decentralized infrastructure for GPU compute and storage started taking real load from AI workloads that didn’t want to depend on hyperscalers.
What’s next
Our shortlist for what to watch in 2025:
- Production-grade agent-to-agent commerce — agents transacting with other agents on rails that humans never touch.
- HIPAA-aware and SOC 2-aware AI offerings becoming table stakes, not a differentiator.
- A formal vocabulary for “the AI agent is also an entity” — somewhere between a user account and a legal person, expressed in verifiable credentials.
- Self-hosted and dedicated AI deployments outgrowing the SaaS-only pattern, especially in regulated verticals.
The framing shift from “blockchain vs AI” to “blockchain + AI” is the story of 2024. The question for 2025 is which products actually ship inside that frame.
Related from Ethora: Ethora AI Agent SDK — embed AI chatbots and agents into your messaging stack with full data ownership.